The Impact of Maternal Employment on Child’s Mental

Health: Evidence from NLSY-Child

Sumanta Mukherjee1

University of Kansas

September 2010

Preliminary and Incomplete

Abstract

An extensive literature has analyzed the effect of a mother’s employment on cognitive outcomes of her children. However, the role of maternal employment in a child’s noncognitive development has received comparatively scant attention. In this paper, data on a panel of children aged four through fifteen are analyzed to explore the effect of maternal employment on a child’s mental health outcomes. Using ordinary least squares and fixed effects estimates, we find that mothers who spend more time at home have children with fewer emotional problems: they score lower on the behavioral problems index; they are also less likely to be frequently unhappy or depressed. In addition, children with mothers spending more time at home are less likely to hurt someone, steal something, or skip school.

JEL Classification: I, J

1 Department of Economics, University of Kansas, 415 Snow Hall, 1460 Jayhawk Blvd., Lawrence, KS 66045. Email: mukherjee@ku.edu

INTRODUCTION

Over the past thirty years, the labor force participation of women with school-age

children has increased significantly. The US Census Bureau reports that the labor force

participation rate of mothers with school-age children increased from 64 percent in 1980

to 78 percent in 1999. The increase in labor force participation was even larger for single

mothers, increasing from 69 percent in 1993 to 83 percent in 1999 (Grogger, 2003). Over

the same period of time, the average annual weeks of employment of mothers increased

from 29.5 to 36.7. At least a part of the recent surge in mothers’ employment is explained

by two distinct, but related changes in the economic environment of low-income

households: a) a tax-credit expansion (EITC) that is designed to pull mothers towards

employment, and b) a welfare reform (TANF) that links welfare receipt with labor market

participation.

While both welfare reforms have achieved the goal of increased labor market

participation, the increased absence of mothers from home has raised concerns about the

potential negative side effects of maternal employment on child development. The

ultimate impact of the mothers’ employment status on children’s outcomes is, however,

not immediately obvious. While an increase in market work may yield benefits through

additional income, a concomitant reduction in mothers’ time at home may have negative

effects on her children.

In this paper, I use mother-child matched data from the National Longitudinal

Survey of the Youth 1979 cohort (NLSY79) to directly test the effect of mothers’

employment on a variety of child outcomes. Using a variety of cognitive and behavioral

outcomes for school-age children, we find that in general maternal employment is

correlated positively with children’s test scores, and negatively correlated with children’s

behavioral problems in a standard OLS estimation framework.

These correlations appear, however, to be mostly driven by unobserved

heterogeneity at the family level: mothers’ labor market status appears to be correlated

with other household specific factors that have a positive influence on child outcomes.

Using fixed effects, I find no significant relationship between maternal employment and

child’s cognitive outcomes as measured by the standardized Mathematics and Reading

Comprehension tests. I do find, however, a large and significant negative effect of

maternal employment on children’s non-cognitive development as indicated by the

Behavior Problems Index (BPI). On average, children with working mothers are

significantly more likely to display distress, anxiety, and similar emotional problems.

The rest of the paper is organized as follows: In Section 2, I describe the existing

literature and attempt to highlight my specific contribution. Section 3 outlines the

empirical strategy. I discuss the data and present some descriptive statistics in Section 4.

The regression results are presented in Section 5. Section 6 concludes.

2. EVIDENCE ON MATERNAL EMPLOYMENT AND CHILD OUTCOMES

An extensive literature analyzes the impact of parental resources on a variety of

child outcomes. Korenman et. al. (1995), for example, uses NLSY data to show that an

increase in current income (1993 dollars) by $10,000 is associated with a small increase

in outcome variables measuring cognitive development and behavioral problems. A

similar increase in permanent income has a somewhat larger positive effect on both

reading ability and behavioral problems.2 Blau (1999) uses the EITC expansion of 1993

to examine the impact of a permanent increase in income on child outcomes for the

NLSY-Child 1979 cohort. He finds that a (nominal) increase in maximum credit from

$953 to $2040, viewed as a permanent increase in income, leads to an increase of at most

1 to 1.5 percent of a standard deviation of the various child outcomes analyzed therein.

He concludes that family income (current and permanent) has a negligible impact on

child development; family background plays a more vital role in determining child

outcomes.

Shea (2000) uses variations in parental ‘luck’ factors (such as job loss experience)

to establish a causal impact of parental income on child outcomes. Echoing the results in

the existing literature, he finds a negligible impact of changes in parental income on

children’s human capital acquisition.3 Dahl and Lochner (2005) utilize the large

expansion in EITC to estimate the causal effect of income on children's math and reading

achievement. Using a fixed-effects instrumental variables strategy they find that a $1,000

increase in income raises math test scores by 2.1 percent and reading test scores by 3.6

percent of a standard deviation. The results are stronger for children from families that

are affected most by the EITC expansion.

The research cited here allows us to draw some broad conclusions regarding

household incomes and child outcomes. Researchers have consistently found that

2 The largest impact of a $10,000 increase in permanent income is 23.5 percent of a standard deviation in reading ability. With fewer controls, the largest effect is a 35.8 percent of a standard deviation improvement in behavioral problems. The effect of a $10,000 increase in income averages 21.6 percent of a standard deviation of the dependent variable across five different outcomes, when income is initially less than half the poverty line, but averages 7.0 percent of a standard deviation beginning from an income level between 1.85 and three times the poverty line. 3 The results of Shea (2000) however indicate that parental income does matter in a sample of low-income households.

increases in income have a minor effect on child outcomes. The EITC expansions, by

providing generous tax breaks, created an alternative natural experiment setting to test the

income effect hypothesis; however the results have still remained surprisingly similar.

A concurrent strand of scholarly work attempts to investigate the importance of

the parental employment in child development.4 Much of the past research on this

question has concentrated on the availability of parental time input in the first three years

of the child. The primary motivation for concentrating on the first few years is derived

from the developmental psychology literature, which emphasizes the effect of early

influences on brain development (Blau and Grossberg, 1992).

The other important characteristic of this literature is the widespread use of some

measure of a child’s cognitive ability as the primary outcome variable. Cognitive

development is typically captured by standardized (or raw) test scores of children of

different age groups. A variety of recent research in this field finds that maternal

employment during the first year of the child’s life has a deleterious effect, while that in

the second or third years has none or some offsetting positive impact (Waldfogel et. al.,

2002; Neidell, 2000). James-Burdumy (2005) uses fixed-effects in determining the effect

of maternal employment on child cognitive outcomes (as measured by PPVT, PIAT Math

and PIAT Reading scores) in the first three years of the child’s life. In particular, she uses

a GMM technique to estimate child and mother fixed-effects regressions. She finds that

the PIAT Math and Reading scores were negatively affected by maternal employment in

the first year. None of the test scores were affected by maternal employment in the

second year. Finally, work in the third year positively affected PIAT Math scores. Ruhm

4 In the absence of detailed time use data, parental employment has often been used as a proxy for the time input in child development (Bernal and Keane 2006).

(2004) finds that adding a more complete set of controls leads to a much stronger

negative impact of maternal employment in the first three years of the child’s life.5

While cognitive skill formation among children has been subject to wide

academic scrutiny, the impediments to a child’s non-cognitive development have

received comparatively scant attention. The literature on investments in children has

traditionally focused on standardized tests scores (such as the PIAT Math and Reading),

with very little emphasis on non-cognitive outcomes.6 The recent literature on human

capital however argues that in addition to cognitive skills (as measured by test scores), a

variety of non-cognitive skills are also very important determinants of subsequent

socioeconomic success (Heckman and Krueger, 2004). The study challenges the

conventional point of view that equates skill with intelligence, adding that “numerous

instances can be cited of people with high IQs who fail to achieve success in life because

they lacked self-discipline and of people with low IQs who succeeded by virtue of

persistence, reliability and self-discipline”. This research therefore cites social skills,

motivation and other non-cognitive skills (in addition to basic intelligence) as key to

labor market success.

5 Indeed there is some literature that finds a completely negative effect of maternal employment for all three years of the child (Han et. al. 2001, Harvey 1999). Liu, Mroz and van der Klaauw (forthcoming) estimate a structural model of mothers’ employment, migration (and hence school) choice, and child outcomes. They also find a negative effect of maternal employment on the cognitive development of children aged 5-15. Bernal (2005) estimates a structural model of employment for married mothers and finds significant negative effects on cognitive achievement of children aged 3-7 years. 6 Bernal and Keane (2006) surveys a large literature on child outcomes (particularly as a consequence of maternal employment). The survey reveals the strong bias in favor of test scores as the primary measure of child outcomes.

Heckman’s concern is reflected in a small but growing literature that attempts to

identify the proximate factors that determine or at least influence the development of

adolescents’ non-cognitive skills. Aizer (2004), for instance, estimates a fixed-effect

linear probability model to determine the impact of child supervision after school on the

probability of risky behavior (skipping school, getting drunk/high, stealing something and

hurting someone). The key explanatory variable is an indicator of whether there is usually

an adult present when the child returns from school.7 The results indicate a lower

probability of risky behavior for children monitored by any adult after school. While this

research provides an interesting analysis of the determinants of non-cognitive

development among school-age children, it does not connect maternal employment (or

unemployment) spells with such outcomes. By using a single indicator variable for child

supervision (which includes post-school monitoring by any adult or day-care), it

potentially misses some interesting effects of the mother’s labor market dynamics on

child development.

My research is distinguished from existing scholarly work in two ways. First, this

paper complements the large literature on very young children by focusing attention on

school-age children. In particular I look at the cognitive outcomes for children aged four

through fifteen. The cognitive outcomes are measured by scores on Peabody Individual

Achievement Tests of Mathematics (PIAT Math) and Reading Comprehension (PIAT

Read) measuring academic achievement of children aged five or older. Second, I

investigate the impact of maternal employment on non-cognitive outcomes as well. In

particular, I assess the effect of maternal employment on Aizer’s index of risky behavior.

7 Children who reported going to day care, or an after-school program, or the home of a relative were considered as being supervised.

In addition, I use the Behavioral Problems Index (BPI) measuring the frequency, range

and type of behavior problems of children aged four and over, as an indicator of a child’s

emotional development.

3. EMPIRICAL FRAMEWORK

The econometric models employed in this analysis are premised on economic

models of the family as a production entity as delineated in Becker (1967, 1981), Becker

and Tomes (1979, 1986) and Leibowitz (1974). These economic models portray

households as productive entities where parents allocate resources (such as income and

time) to maximize an objective function that includes the health and development of

children as arguments. I use this setup to specify a variety of child outcomes as linear

functions of parental resources and a vector of controls in the following form:

jtjtjtjtjt HEXy εγβα +++= (1),

where jty is the jth child’s outcome in year t, E measures the fraction of weeks mother

was employed since last interview (maternal work at the ‘extensive’ margin), H

represents the number of hours worked per week (as a measure of the ‘intensive’ margin

of employment), X is a vector of controls, ε is the error term, and βα , andγ are

parameters. The basic models are estimated by ordinary least squares (OLS) with

clustered standard errors.

To account for the potential unobservable heterogeneity, I employ a mother fixed-

effect to control for unobserved time-invariant factors (such as genetic or environmental

influences) specific to the mother and the household. The fixed-effect model is robust to

the endogeneity of any explanatory variables provided that the fixed effect is the only

source of correlation between the regressors and the error term. The mother fixed-effect

model for the ith family is specified as follows:

ijtijijtijtijtijt HEXy εθγβα ++++= (2),

where ijθ represents the mother fixed-effect for the ith mother or family. The error term

can be decomposed into three components: ijtjiijt υηϕε ++= , where ϕ and η represent

the family-specific and individual-specific effects respectively. Assuming that

unobservable maternal traits are sibling-invariant ( iij θθ = ), and operate only through the

family-specific effect ϕ, one can take first-differences with respect to families to obtain:

ijijijijij HEXy εγβα ∆+∆+∆+∆=∆ (3).

First differencing within families therefore eliminates any observed or unobserved

variables that do not vary (over time) within a family. The model is identified under the

assumption that the differences in exposure to maternal employment across siblings are

exogenous to the children’s development.

4. DATA AND DESCRIPTIVE STATISTICS

The National Longitudinal Survey of the Youth has regularly interviewed a

national sample of individuals who were aged 14 to 21 as of December 31, 1978. The

sample over-represents Blacks, Hispanics, low-income Whites and military personnel. In

each even numbered year since 1986, the children born to the female participants of the

NLSY sample have been administered a set of assessments of cognitive, social, emotional

and physical development together with the quality of home environment. The data used

in the paper are from the first eleven waves of the NLSY Child files collected over the

period 1986 to 2006. My sample includes mother-child matched data for children in their

school years (i.e. aged about five through eighteen) for whom measures of cognitive and

non-cognitive development are available. The particular choice of age group reflects the

hypothesis that non-cognitive skill acquisition is most affected by parental time in this

age interval.

The primary outcome variables in the second essay are measures of cognitive and

non-cognitive development of children. The cognitive outcomes are captured by the

scores on the Peabody Individual Assessment Tests of Mathematics (PIATMATH) and

Reading Recognition (PIATREAD) that indicate academic achievement for children aged

five and above. Following Aizer (2004), I use her index of antisocial or risky child

behavior as a measure of the child’s non-cognitive development. The index is based on

the following specific behaviors: skipping school, getting drunk/high, stealing something

and hurting someone bad enough to need a doctor. In addition, the lack of maternal time

could lead to nontrivial emotional problems for children and adolescents. Therefore I use

the Behavioral Problems Index (BPI) measuring the frequency, range and type of

behavior problems of children aged four and over, as an indicator of a child’s emotional

development.

The two primary explanatory variables measure the mothers’ labor market

commitments. First, I use the fraction of weeks worked since last interview as a measure

of maternal work at the extensive margin. Second, I use the hours of work per week

worked as a measure of work at the intensive margin.

A variety of supplementary ‘home’ inputs tend to influence a child’s

development. The ‘Home Inventory’ variables reported in the NLSY79 are used to

summarize the quantity and quality of these supplementary inputs available in the child’s

home. This measure is derived from a series of child-age-specific questions and

interviewer observations on the home environment and the nature of mother-child

interactions. In addition, I use a number of core maternal background regressors that are

intended to capture the education, demographic characteristics, income and location of

the mother.8

Along with income, a household’s size can often determine the amount of care

(specifically, time) allocated to an individual child. In some cases, the birth order and

gender of an individual child influences his/her outcomes. Race can often play a

significant role in determining the nature and extent of family ties and informal child

supervision possibilities. Therefore I add some other controls that capture the child’s

gender, race, birth order and number of siblings. Finally I also incorporate an indicator

for the child’s living arrangement, which is equal to 1 if the child stays with the mother

and 0 otherwise. Table 1 outlines the descriptive statistics for the variables used in this

essay.

8 For mother’s education, we use information on the highest grade completed by mother. I have not included mothers’ AFQT test results because there is little consensus in the literature on what the score really measures.

5. RESULTS

The main regression results for four different outcome variables are presented in

Tables 2 to 4. Tables 2 and 3 present estimates of the effect of maternal employment on

PIAT Math and Reading scores. The OLS results in both tables indicate a strong impact

of maternal work (on both the ‘extensive’ and ‘intensive’ margins) on cognitive

achievement. However the effect ceases to be significant when I use mother fixed-effects.

The basic positive correlation appears to be mostly driven by mother and household-

specific factors that are positively correlated with both income and employment and thus

generate a classical omitted variable bias in the OLS estimates. Even with household

fixed effects, however, the measure of home inputs turns out a highly significant

determinant of child’s cognitive outcomes and is robust to changes in specification or

sample. These results strongly highlight the importance of home environment on a child’s

cognitive development. The results reported in Tables 2 (Math) and 3 (Reading and

Comprehension) look overall very similar. Overall maternal employment appears to be

slightly more detrimental for boys (column 3 of Tables 2 and 3), even though the estimate

is significantly different from zero only for the reading and comprehension score.

Table 4 presents estimates of the effect of maternal employment on a child’s

emotional well-being as reflected by the BPI score. While the OLS results do not show

significant results, all the specifications with mother fixed-effects indicate that maternal

employment at the ‘extensive’ margin has a highly significant effect on behavioral

problems. In particular, it appears that a larger number of weeks of employment for a

mother significantly exacerbated behavioral problems for her children. These results

appear to be very robust across sampling strata; the magnitudes and significance of the

coefficients change very little when I restrict the sample to boys (column 3), African

Americans (column 4) or Hispanics (column 5). The number of hours worked per week,

on the other hand, has little impact on the BPI score. Home inputs turn out to be

important determinants of a child’s emotional well-being, with a higher quality of home

environment significantly reducing behavioral problems. Two other interesting points

emerge. In all the specifications, first and second born children do much better than their

younger siblings in dealing with behavioral problems. Finally, girls fare significantly

better than boys on BPI score across all specifications and samples.

In Table 5, I test the effect of maternal employment on a behavioral problem

index suggested by Aizer (2004). The dependent variable is an indicator, which is coded

to 1 if the child has either stolen, skipped school, got drunk or hurt anybody. The results

are mixed, and somewhat hard to interpret. The OLS results imply that more hours per

week lead to higher scores on child risky behaviors, but more participation in the labor

market is associated with lower behavioral problems. This positive effect of labor force

participation disappears in the full sample once I control for mother fixed effects. It does,

however, persist in the male sample. The interpretation of this finding is not

straightforward. Given the self-reported nature of the behavioral problems index, one

might argue that boys with working mothers tend to report less of their bad deeds.

6. CONCLUSIONS

With rapidly increasing female labor force participation over the last few decades,

a lot of research has been dedicated to identifying the causal effect of maternal work

during early childhood on subsequent child development; relatively little attention has

been given to the effect of maternal employment during the later stages of childhood (as

well as early adulthood). In this paper I have used longitudinal data from the US to test

whether a mother’s attachment to the labor market during the ages 5-14 significantly

affects a child’s cognitive and non-cognitive development. While we find that maternal

employment has no effect on the cognitive development of children, we find a strong and

highly significant relationship between maternal employment and the frequency and

severity of children’s behavioral and psychological problems as summarized in the

Behavioral Problems Index (BPI). This effect appears to be consistent across gender and

ethnicities.

The results in this paper not only highlight the important impact maternal

employment during childhood and early adulthood may have on child development; they

also underline the very differential effects parental choices may have on the

psychological and mental well-being of their children. While parental employment is

likely to have only an indirect positive effect on children’s school performance, it appears

to have a rather important negative effect on the child’s emotional well-being. Even

though the magnitude of these effects may be harder to quantify than the consequences of

poor performance in school, they should clearly not be neglected from a broader welfare

perspective.

REFERENCES

Aizer, A. (2004). “Home Alone: Supervision after School and Child Behavior”, Journal of Public Economics, 88, 1835 – 1848. Becker, G. (1967). “Human Capital and the Personal Distribution of Income: An Analytic Approach”, Woytinski Lecture, no. 1, University of Michigan, Institute of Public Administration, Ann Arbor. Becker, G. (1981). A Treatise on the Family. Cambridge, MA: Harvard University Press. Becker, G. and N. Tomes (1979). “ An Equilibrium Theory of the Distribution of Income and Intergenerational Mobility”, Journal of Political Economy, 87(6), pp. 1153-89. Becker, G. and N. Tomes (1986). “Human Capital and the Rise and Fall of Families”, Journal of Labor Economics, 4 (3, Part 2), pp. S1-39. Bernal, R. (2005). “The Effect of Maternal Employment and Child Care on Children’s Cognitive Development”, mimeo, Northwestern University. Bernal, R. and M. Keane (2006). “Child Care Choices and Children’s Cognitive Achievement: The Case of Single Mothers”, mimeo, Northwestern University. Blau, D. (1999). “The Effect of Income on Child Development”, Review of Economics and Statistics, 81(2), 261 – 276. Blau, F. and A. Grossberg (1992). “Maternal Labor Supply and Children’s Cognitive Development”, Review of Economics and Statistics, 74(3), 474 – 481. Dahl, G. and L. Lochner (2005). “The Impact of Family Income on Child Achievement”, NBER working paper # 11279. Grogger, J. and L. Karoly (2003). Welfare Reform. Cambridge, MA: Harvard University Press.

Han, W., J. Waldfogel and J. Brooks-Gunn (2001). “The Effects of Early Maternal Employment on Later Cognitive and Behavioral Outcomes”, Journal of Marriage and

Family, 63, 336 – 354. Harvey, E. (1999). “Short Term and Long Term Effects of Early Parental Employment on Children of the National Longitudinal Survey of Youth”, Developmental Psychology, 35(2), 445 – 459. Haveman, R. and B. Wolfe (1995). “The Determinants of Children’s Attainments: A Review of Methods and Findings”, Journal of Economic Literature, 33(4), 1829 – 1878.

Heckman, J. and A. Krueger (2003). Inequality in America: What Role for Human

Capital Policies. Cambridge: MIT Press. James-Burdumy, S. (2005). “The Effect of Maternal Labor Force Participation on Child Development”, Journal of Labor Economics, 23(1), 177 – 211. Korenman, S., J. Miller and J. Sjaastad (1995). “Long-Term Poverty and Child Development in the United States: Results From the NLSY”, Child and Youth Service Review, 17(12), 127-55. Lang, K. and J. Zagorsky (2001). “Does Growing Up with a Parent Absent Really Hurt?” Journal of Human Resources, 36(2), 253 – 273. Liu, H., T. Mroz and W. van der Klaauw (forthcoming). “Maternal Employment, Migration, and Child Development”, Journal of Econometrics. Neidell, M. (2000). “Early Parental Time Investments in Children’s Human Capital Development: Effects of Time in the First Year on Cognitive and Non-Cognitive Outcomes”, mimeo, UCLA. Ruhm, C. (2004). “Parental Employment and Child Cognitive Development”, Journal of Human Resources, 39(1), 155 – 192. Shea, J. (2000). “Does Parents’ Money Matter?” Journal of Public Economics, 77(2), 155 – 184. Waldfogel, J., W. Han and J. Brooks-Gunn (2002). “The Effects of Early Maternal Employment on Child Cognitive Development”, Demography, 39(2), 369 – 392.

Table 1: Descriptive Statistics from NLSY-Child 1979

Variable Obs Mean Std. Dev. Min Max

PIAT Math Raw Score 22894 100.20 13.91 0 135

Behavioral Problems Index Score 22894 64.23 62.39 0 280

Age 22894 9.57 2.81 4.916667 18.08333

Age mother 22894 33.99 5.08 21 47

African American 22894 0.30 0.46 0 1

First born child 22894 0.44 0.50 0 1

Hispanic 22894 0.20 0.40 0 1

Lives with mother 22894 0.98 0.14 0 1

Sibling under age 2 or younger 22894 0.17 0.37 0 1

Mother age at birth 22894 24.40 5.00 11 40

Number of siblings 22894 2.61 1.19 0 9

Home inventory score 22894 974 157 80 1333

Mother education 22894 12.57 2.31 0 20

Family income 22894 43633 68783 0 974100

Urban area 22894 0.75 0.44 0 1

Mother married 22894 0.62 0.49 0 1

Mother single 22894 0.14 0.35 0 1

Fraction employed 22894 0.63 0.43 0 1

Fraction unemployed 22894 0.04 0.14 0 1

Table 2: PIAT Math Scores and labor market participation

Dependent Variable

Standardized PIAT Math Score

(1

) (2

) (3

) (4

) (5

)

Fra

ctio

n o

f yea

r em

plo

yed

1.2

88***

-0.0

464

-0.2

70

0.0

454

0.6

30

(0

.45)

(0.4

3)

(0.6

0)

(0.7

7)

(0.9

3)

Aver

age

hours

per

wee

k

-0.0

440***

-0.0

109

-0.0

000362

-0.0

213

-0.0

324*

(0

.010)

(0.0

095)

(0.0

13)

(0.0

18)

(0.0

20)

Num

ber

of si

blings

-0.5

04***

-0.2

96

-0.3

00

-0.3

40

-0.6

98

(0

.17)

(0.2

5)

(0.4

1)

(0.3

9)

(0.4

9)

Hom

e in

ven

tory

sco

re

0.0

115***

0.0

0494***

0.0

0424***

0.0

0576***

0.0

0307

(0

.00098)

(0.0

0092)

(0.0

013)

(0.0

015)

(0.0

020)

Sam

ple

A

ll

All

Boys only

B

lack

H

ispan

ic

Spec

ific

atio

n

OLS

FE

FE

FE

FE

Obse

rvat

ions

17468

17468

17468

17468

17468

R-s

quar

ed

0.1

8

-63.0

9

0.5

5

0.6

4

0.5

0

Robust sta

ndar

d e

rrors

in p

aren

thes

es a

re c

luster

ed a

t th

e fa

mily lev

el.

*** p

<0.0

1, ** p

<0.0

5, * p

<0.1

A

ll spec

ific

atio

ns co

ntrol fo

r se

x, birth

ord

er, et

hnic

ity, livin

g w

ith m

oth

er, m

oth

er’s

mar

ital

sta

tus an

d e

duca

tion, an

d u

rban

res

iden

ce.

Table 3: PIAT Reading and Comprehension Score and Labor Market Participation

Dependent Variable

Standardized PIAT Reading and Comprehension Score

(1

) (2

) (3

) (4

) (5

)

Fra

ctio

n o

f yea

r em

plo

yed

1.7

04***

0.0

267

-1.2

01*

0.4

28

-0.2

40

(0

.44)

(0.4

6)

(0.6

5)

(0.8

0)

(0.9

9)

Aver

age

hours

per

wee

k

-0.0

451***

-0.0

131

-0.0

0688

-0.0

163

-0.0

0785

(0

.010)

(0.0

10)

(0.0

16)

(0.0

21)

(0.0

26)

Num

ber

of si

blings

-0.5

46***

-0.1

83

0.1

94

-0.3

54

-0.3

77

(0

.16)

(0.2

6)

(0.3

6)

(0.4

2)

(0.6

3)

Hom

e in

ven

tory

sco

re

0.0

114***

0.0

0434***

0.0

0301**

0.0

0468***

0.0

0276

(0

.0010)

(0.0

010)

(0.0

015)

(0.0

016)

(0.0

023)

Sam

ple

A

ll

All

Boys only

B

lack

H

ispan

ic

Spec

ific

atio

n

OLS

FE

FE

FE

FE

Obse

rvat

ions

14828

14828

7330

4432

2811

R-s

quar

ed

0.2

3

0.5

9

0.6

8

0.6

1

0.5

3

Robust sta

ndar

d e

rrors

in p

aren

thes

es a

re c

luster

ed a

t th

e fa

mily lev

el.

*** p

<0.0

1, ** p

<0.0

5, * p

<0.1

A

ll spec

ific

atio

ns co

ntrol fo

r se

x, birth

ord

er, et

hnic

ity, livin

g w

ith m

oth

er, m

oth

er’s

mar

ital

sta

tus an

d e

duca

tion, an

d u

rban

res

iden

ce.

Table 4: Behavior Problems Index (BPI) and Labor Market Participation

Dependent Variable

Behavioral Problems Index (BPI) Raw Score

(1

) (2

) (3

) (4

) (5

)

Fra

ctio

n o

f yea

r em

plo

yed

2.6

58

12.1

8***

13.1

3***

13.4

8***

12.2

7**

(2

.50)

(2.5

6)

(3.4

9)

(4.7

1)

(5.6

9)

Aver

age

hours

per

wee

k

0.0

0496

-0.0

820

-0.1

18

-0.0

762

-0.2

27*

(0

.054)

(0.0

55)

(0.0

78)

(0.1

1)

(0.1

2)

Num

ber

of si

blings

-2.5

76***

4.4

68***

4.1

17**

4.5

37*

7.2

38***

(0

.83)

(1.2

5)

(1.7

3)

(2.3

9)

(2.3

6)

Hom

e in

ven

tory

sco

re

-0.0

846***

-0.0

347***

-0.0

219***

-0.0

337***

-0.0

387***

(0

.0052)

(0.0

052)

(0.0

075)

(0.0

092)

(0.0

11)

Sam

ple

A

ll

All

Boys only

B

lack

H

ispan

ic

Spec

ific

atio

n

OLS

FE

FE

FE

FE

Obse

rvat

ions

17468

17468

8755

5134

3355

R-s

quar

ed

0.0

7

0.5

3

0.6

0

0.5

1

0.5

2

Robust sta

ndar

d e

rrors

in p

aren

thes

es a

re c

luster

ed a

t th

e fa

mily lev

el.

*** p

<0.0

1, ** p

<0.0

5, * p

<0.1

A

ll spec

ific

atio

ns co

ntrol fo

r se

x, birth

ord

er, et

hnic

ity, livin

g w

ith m

oth

er, m

oth

er’s

mar

ital

sta

tus an

d e

duca

tion, an

d u

rban

res

iden

ce.

Table 5: Aizer Behavioral Index and Labor Market Participation

Dependent Variable

Aizer Behavioral Problem Index

(1

) (2

) (3

) (4

) (5

)

Fra

ctio

n o

f yea

r em

plo

yed

-0

.0508**

-0.0

117

-0.1

33**

-0.0

643

-0.0

172

(0

.026)

(0.0

40)

(0.0

64)

(0.0

73)

(0.0

84)

Aver

age

hours

per

wee

k

0.0

0114**

-0.0

0114

-0.0

00712

-0.0

0166

-0.0

0126

(0

.00048)

(0.0

0086)

(0.0

012)

(0.0

018)

(0.0

019)

Num

ber

of si

blings

0.0

126**

-0.0

518***

-0.0

695**

-0.0

971***

-0.0

0345

(0

.0064)

(0.0

20)

(0.0

34)

(0.0

32)

(0.0

46)

Hom

e in

ven

tory

sco

re

-0.0

00175***

-0.0

00185**

-0.0

00146

-0.0

00249*

-0.0

00120

(0

.000051)

(0.0

00074)

(0.0

0012)

(0.0

0013)

(0.0

0015)

Sam

ple

A

ll

All

Boys only

B

lack

H

ispan

ic

Spec

ific

atio

n

OLS

FE

FE

FE

FE

Obse

rvat

ions

6011

6011

2966

1785

1204

R-s

quar

ed

0.0

7

0.4

9

0.6

2

0.4

9

0.5

1

Robust sta

ndar

d e

rrors

in p

aren

thes

es a

re c

luster

ed a

t th

e fa

mily lev

el.

*** p

<0.0

1, ** p

<0.0

5, * p

<0.1

A

ll spec

ific

atio

ns co

ntrol fo

r se

x, birth

ord

er, et

hnic

ity, livin

g w

ith m

oth

er, m

oth

er’s

mar

ital

sta

tus an

d e

duca

tion, an

d u

rban

res

iden

ce.

The

dep

enden

t var

iable

is co

ded

to 1

, if c

hild rep

orts to

hav

e ei

ther

, stole

n, got dru

nk, sk

ipped

sch

ool or hurt som

ebody.